New methods for calculations of the lowest eigenvalues of the real symmetric generalized eigenvalue problem (Q1578093)

The author proposes a new method to find the smallest eigenvalue and the corresponding eigenvector of the generalized eigenvalue problem \(AX=\lambda BX\), where \(A\) and \(B\) are real symmetric matrices, and \(B\) is also positive definite. The method proceeds iteratively, and at each step it solves the problem in a subspace, then enlarges the subspace by means of the solution of the Newton secant equation, with the data from the subspace solution, when Newton's method is applied to find a solution that minimizes \[ \rho(X)= \frac{(X,AX)}{(X,BX)}. \] The method and some variants are tested in various examples and the convergence results are similar to those obtained by means of the Jacobi-Davidson method.